Lowermost tropospheric ozone observed from space by
multispectral synergism of IASI (IR) and GOME-2 (UV)
J. Cuesta1, M. Eremenko1, X. Liu2, G. Dufour1, Z. Cai3, M. Höpfner4, T. von Clarmann4, P. Sellitto1, G. Foret1, B. Gaubert1, M. Beekmann1,
J. Orphal4, K. Chance2 , R. Spurr5 and J.-M. Flaud1
(1) (2) (3) (4) (5)
Scientific motivation
Tropospheric ozone is a major atmospheric pollutant
� It directly affects public health, aggravating respiratory diseases and increasing premature mortality
� It damages ecosystems and reduces plants growth
� Emissions of precursors reduced significantly in the last decades in regions as Europe, but it has not been reflected in O3 concentrations
Ø Monitoring tropospheric ozone is mandatory
➙ Only satellite observation can measure its daily and seasonal evolution at the regional and global scales
q How to observe O3 in the lowermost troposphere from space ?
q Which UV/TIR spectrocopic data are coherent enough for synergism?
We propose a new multispectral synergism of IASI (IR) and GOME-2 (UV)
VLIDORT
Multi-spectral retrieval of Lowermost Tropospheric Ozone
Effective albedo, cloud fraction and
calibration
Atmospheric conditions T(z), P(z), H2O(z)Surface properties ε, ω,Ts
Spectroscopic data TIR & UV
Tikhonov-Phillips regularizationfor UV+TIRCalibration
and H2O(z)
KOPRA
TIR Radiance spectrum
UV Reflectancespectrum
Validation withOzone sondes
GOME-2IASI
IASI+GOME-2: a new joint inversion of collocated IR and UV spectra using Tikhonov-Phillips constraints
[Eremenko et al., 2008] [Cai et al., 2012]
[Cuesta et al., 2013, ACPD]
Brion et al., 1993 UV cross sections
HITRAN2004 TIR lines
parameters
Slit functions, « Soft »
recalibration
980-1070 cm-1 290-345 nm
apodisation, offsets
Simultaneous fit of IR and UV spectra
Adjusting a unique Ozone profile and
instrumental parameters
UV Reflectances IR Radiances
∼9.5%
∼0.25 % ∼0.39 %
1. Multiple spectral fitting
Systematic erros (also in UV only)
2. Multispectral sensitivity: Averaging kernels IASI (IR) GOME-2 (UV) IASI+GOME-2
Higher sensitivity for lower
layers
Two semi-independent tropospheric
partial columns
2. Multispectral sensitivity: Degrees of freedom in the Lowermost Troposphere (up to 3 km asl)
IASI (IR) GOME-2 (UV) IASI+GOME-2
0.25 DOFs over land 0.15 DOFs over ocean <0.10 DOFs
0.35 DOFs over land 0.25 DOFs over ocean
DOFIASI + 40%
2. Multispectral sensitivity: Height of maximum sensitivity in the Lowermost Troposphere
IASI (IR) GOME-2 (UV) IASI+GOME-2
3 km agl over land 4.3 km agl over ocean 3.7 km agl
2.2 km agl over land 3.5 km agl over ocean
HIASI - 800 m
3. Validation against ozonesondes : IASI+GOME-2 vs. Smoothed Ozonesondes
(HITRAN2004 + Brion et al., 1993)
IAS
I+G
OM
E-2
IAS
I+G
OM
E-2
119 sondes during the summer of 2009 (57 days), from 10 stations over Europe
Lower Troposphere (LT) (up to 6 km asl)
Lowermost Troposphere (LMT) (up to 3 km asl)
R=0.70 Bias=-4.1% RMS=17%
R=0.87 Bias=-2.1% RMS=17%
Good agreement: Low mean bias and reasonable RMS differences for the LT and LMT. Correlation is particularly good for the LMT
IAS
I+G
OM
E-2
IAS
I+G
OM
E-2
Lower Troposphere (LT) (up to 6 km asl)
Lowermost Troposphere (LMT) (up to 3 km asl)
R=0.58 Bias=-3.2% RMS=20%
R=0.75 Bias=-0.7% RMS=24%
Direct comparisons: LMT retrieval also shows very low mean bias and good correlation
3. Validation against ozonesondes : IASI+GOME-2 vs. Raw Ozonesondes
(HITRAN2004 + Brion et al., 1993) 119 sondes during the summer of 2009 (57 days), from 10 stations over Europe
4. Spectroscopic coherence UV vs TIR ?
Similar results with UV cross sections -5%���
Aberrant features in the troposphere
IAS
I+G
OM
E-2
w
ith
HIT
RA
N2000
IAS
I+G
OM
E-2
w
ith
HIT
RA
N2000
IAS
I+G
OM
E-2
w
ith
HIT
RA
N2000
IAS
I+G
OM
E-2
w
ith
HIT
RA
N2000
4. Spectroscopic coherence : using HITRAN 2000 + Brion et al. 1993 ?
Strong negative
bias, lower
correlation and higher
RMS
Among the tested data, the best spectroscopic parameters for IASI+GOME-2 is HITRAN2004+BRION1993
4. Lower tropospheric ozone observations : IASI+GOME-2 vs. IASI and GOME-2
IASI (IR) GOME-2 (UV) IASI+GOME-2
19 August 2009
Ozone plumes depicted by both IASI and IASI+GOME-2 Only observed by
IASI+GOME-2
4. Lower tropospheric ozone observations : IASI+GOME-2 vs. IASI and GOME-2
IASI (IR) GOME-2 (UV) IASI+GOME-2
20 August 2009
Only observed by IASI+GOME-2 Ozone plumes depicted by both IASI and IASI+GOME-2
5. Lowermost tropospheric ozone : IASI+GOME-2 observations vs. CHIMERE model
IAS
I+G
OM
E-2
LM
T
19 August 2009
CH
IME
RE
* A
VK
LM
T
RA
W C
HIM
ER
E
LMT
R
AW
CH
IME
RE
3 –
6 k
m a
sl
Only below 3 km asl
Also above 3 km asl
5. Lowermost tropospheric ozone : IASI+GOME-2 observations vs. CHIMERE model 20 August 2009
RA
W C
HIM
ER
E
LMT
R
AW
CH
IME
RE
3 –
6 k
m a
sl
IAS
I+G
OM
E-2
LM
T
CH
IME
RE
* A
VK
LM
T
Summary v We have developed a new multispectral method to observe
tropospheric ozone from space: “IASI+GOME-2” ü It simultaneously fits UV and TIR spectra
ü It enables a mean enhancement of sensitivity of 40% to ozone in the LMT
ü It uses HITRAN2004+BRION1993 ➙ Best for IASI+GOME-2 (note that spectra are recalibrated within the method)
ü It shows a good agreement with ozonesondes, both smoothed by the retrieval AVK and raw measurements
ü Only IASI+GOME-2 depicts lowermost tropospheric ozone plumes (as shown by CHIMERE) ➙ Air Quality studies
� Future developments will focus on § The contribution of the Chappuis band in the visible, towards a
three band TIR+UV+VIS ozone retrieval § Radiative effects of aerosols on ozone retrievals
Acknowledgements
21/09/12 Rapport d'archives -‐‑ WOUDC -‐‑ [Service météorologique du Canada]
1/1www.woudc.org/data/MetaQuery/MetaReport_f.cfm?ARC=A&S=A&C=C,OzoneSonde,1.0&L=G,35,…
�0HWHRURORJLFDO�6HUYLFH�RI�&DQDGD :RUOG�0HWHRURORJLFDO�2UJDQL]DWLRQ�
�:28'&��$FWLYH�'DWD�LQ�$UFKLYH�6XPPDU\�5HSRUW������UHFRUGV�IRXQG��
&LWDWLRQ��$(0(7���&+0,�35���':'�02+S���':'�02/���.10,���0(��0HWHR6ZLVV���3,0:0���50,%���7606���8.02��:RUOG�2]RQH�DQG
8OWUDYLROHW�5DGLDWLRQ�'DWD�&HQWUH��:28'&��>'DWD@��5HWULHYHG�6HSWHPEHU����������IURP�KWWS���ZZZ�ZRXGF�RUJ�
5HSRUW�'DWH�������������&DWHJRU\���2]RQH6RQGH���/HYHO����
0LQLPXP�/DWLWXGH������GHFLPDO�GHJUHHV0D[LPXP�/DWLWXGH������GHFLPDO�GHJUHHV
0LQLPXP�/RQJLWXGH�������GHFLPDO�GHJUHHV0D[LPXP�/RQJLWXGH������GHFLPDO�GHJUHHV
%HJLQQLQJ�<HDU�������(QGLQJ�<HDU�������
,QIRUPDWLRQ�RQ�UHWUHLYLQJ�GDWD�IURP�WKH�:28'&�$UFKLYH
6KRZ�3ODWIRUP�ORFDWLRQV�IURP�WKLV�UHSRUW�LQ�0DS�9LHZ
6DYH�WKLV�UHSRUW�GDWD�WR�&69
Data
Archive
Category Platform
ID
Platform Name Country WMO
Region
Latitude Longitude Earliest
Data
Latest Data Last
Updated
Files
Link OzoneSonde - Level 1.0 STN 43 LERWICK GBR VI 60.1315 -1.183 2009-01-07 2009-12-30 2010-11-16 52
Link OzoneSonde - Level 1.0 STN 53 UCCLE BEL VI 50.8 4.35 2009-01-07 2009-12-23 2011-01-19 141
Link OzoneSonde - Level 1.0 STN 99 HOHENPEISSENBERG DEU VI 47.8 11.02 2009-01-02 2009-12-30 2010-07-09 127
Link OzoneSonde - Level 1.0 STN 156 PAYERNE CHE VI 46.49 6.57 2009-01-02 2009-12-30 2010-03-17 154
Link OzoneSonde - Level 1.0 STN 174 LINDENBERG DEU VI 52.21 14.12 2009-01-07 2009-12-30 2010-01-25 50
Link OzoneSonde - Level 1.0 STN 221 LEGIONOWO POL VI 52.4 20.967 2009-01-08 2009-12-30 2010-04-26 43
Link OzoneSonde - Level 1.0 STN 242 PRAHA CZE VI 50.02 14.45 2009-01-02 2009-04-28 2009-05-20 45
Link OzoneSonde - Level 1.0 STN 308 MADRID / BARAJAS ESP VI 40.46 -3.65 2009-01-09 2009-12-23 2010-01-18 48
Link OzoneSonde - Level 1.0 STN 316 DE BILT NLD VI 52.1 5.18 2009-01-08 2009-12-31 2010-03-01 46
Link OzoneSonde - Level 1.0 STN 318 VALENTIA OBSERVATORY IRL VI 51.93 -10.25 2009-01-02 2009-12-30 2010-01-18 54
Link OzoneSonde - Level 1.0 STN 348 ANKARA TUR VI 39.95 32.883 2009-01-14 2009-11-18 2010-06-16 17
�7RWDO�)LOHV�IRU�6HDUFK�&ULWHULD������
21/09/12 Rapport d'archives -‐‑ WOUDC -‐‑ [Service météorologique du Canada]
1/1www.woudc.org/data/MetaQuery/MetaReport_f.cfm?ARC=A&S=A&C=C,OzoneSonde,1.0&L=G,35,…
�0HWHRURORJLFDO�6HUYLFH�RI�&DQDGD :RUOG�0HWHRURORJLFDO�2UJDQL]DWLRQ�
�:28'&��$FWLYH�'DWD�LQ�$UFKLYH�6XPPDU\�5HSRUW������UHFRUGV�IRXQG��
&LWDWLRQ��$(0(7���&+0,�35���':'�02+S���':'�02/���.10,���0(��0HWHR6ZLVV���3,0:0���50,%���7606���8.02��:RUOG�2]RQH�DQG
8OWUDYLROHW�5DGLDWLRQ�'DWD�&HQWUH��:28'&��>'DWD@��5HWULHYHG�6HSWHPEHU����������IURP�KWWS���ZZZ�ZRXGF�RUJ�
5HSRUW�'DWH�������������&DWHJRU\���2]RQH6RQGH���/HYHO����
0LQLPXP�/DWLWXGH������GHFLPDO�GHJUHHV0D[LPXP�/DWLWXGH������GHFLPDO�GHJUHHV
0LQLPXP�/RQJLWXGH�������GHFLPDO�GHJUHHV0D[LPXP�/RQJLWXGH������GHFLPDO�GHJUHHV
%HJLQQLQJ�<HDU�������(QGLQJ�<HDU�������
,QIRUPDWLRQ�RQ�UHWUHLYLQJ�GDWD�IURP�WKH�:28'&�$UFKLYH
6KRZ�3ODWIRUP�ORFDWLRQV�IURP�WKLV�UHSRUW�LQ�0DS�9LHZ
6DYH�WKLV�UHSRUW�GDWD�WR�&69
Data
Archive
Category Platform
ID
Platform Name Country WMO
Region
Latitude Longitude Earliest
Data
Latest Data Last
Updated
Files
Link OzoneSonde - Level 1.0 STN 43 LERWICK GBR VI 60.1315 -1.183 2009-01-07 2009-12-30 2010-11-16 52
Link OzoneSonde - Level 1.0 STN 53 UCCLE BEL VI 50.8 4.35 2009-01-07 2009-12-23 2011-01-19 141
Link OzoneSonde - Level 1.0 STN 99 HOHENPEISSENBERG DEU VI 47.8 11.02 2009-01-02 2009-12-30 2010-07-09 127
Link OzoneSonde - Level 1.0 STN 156 PAYERNE CHE VI 46.49 6.57 2009-01-02 2009-12-30 2010-03-17 154
Link OzoneSonde - Level 1.0 STN 174 LINDENBERG DEU VI 52.21 14.12 2009-01-07 2009-12-30 2010-01-25 50
Link OzoneSonde - Level 1.0 STN 221 LEGIONOWO POL VI 52.4 20.967 2009-01-08 2009-12-30 2010-04-26 43
Link OzoneSonde - Level 1.0 STN 242 PRAHA CZE VI 50.02 14.45 2009-01-02 2009-04-28 2009-05-20 45
Link OzoneSonde - Level 1.0 STN 308 MADRID / BARAJAS ESP VI 40.46 -3.65 2009-01-09 2009-12-23 2010-01-18 48
Link OzoneSonde - Level 1.0 STN 316 DE BILT NLD VI 52.1 5.18 2009-01-08 2009-12-31 2010-03-01 46
Link OzoneSonde - Level 1.0 STN 318 VALENTIA OBSERVATORY IRL VI 51.93 -10.25 2009-01-02 2009-12-30 2010-01-18 54
Link OzoneSonde - Level 1.0 STN 348 ANKARA TUR VI 39.95 32.883 2009-01-14 2009-11-18 2010-06-16 17
�7RWDO�)LOHV�IRU�6HDUFK�&ULWHULD������
27
Table 1. Sensitivity of the multispectral ozone retrieval IASI+GOME-2 (UV+TIR) and 1
comparison with IASI (TIR) and GOME-2 (UV) only retrievals. Degrees of freedom for 2
signal and the height of maximum sensitivity (km agl) are given for different atmospheric 3
partial columns (LMT, LT, TROPO and TOTAL are respectively from the surface up to 3, 6, 4
12 and 60 km asl). Statistics are given in mean values and standard deviations (with respect to 5
the average, in parenthesis) for all pixels over land and over ocean on 19-20 August 2009 for 6
the region shown in Fig. 4. In order to compare partial columns of similar depths, we consider 7
only land pixels with ground altitudes below 300 m asl. For these pixels, the average thermal 8
contrast is 4.4 ± 2.4 K over land and 0.9 ± 1.5 K over ocean. 9
10
IASI+GOME-2 IASI GOME-2
Atmospheric
column Land Ocean Land Ocean Land Ocean
Degrees of
freedom for
signal
colDOF
LMT 0.34
(±0.04)
0.23
(±0.04)
0.24
(±0.03)
0.16
(±0.04)
0.08
(±0.01)
0.08
(±0.02)
LT 0.75
(±0.05)
0.64
(±0.05)
0.62
(±0.03)
0.52
(±0.04)
0.25
(±0.03)
0.24
(±0.04)
TROPO 1.72
(±0.07)
1.51
(±0.12)
1.52
(±0.06)
1.34
(±0.09)
0.67
(±0.04)
0.65
(±0.05)
TOTAL 5.20
(±0.12)
4.92
(±0.19)
3.43
(±0.10)
3.20
(0.10)
3.41
(±0.10)
3.32
(±0.12)
Height of
maximum
sensitivity maxcolH
LMT 2.20
(±0.50)
3.48
(±0.56)
3.02
(±0.67)
4.26
(±0.51)
3.68
(±0.50)
3.64
(±0.65)
LT 2.45
(±0.55)
3.83
(±0.57)
3.21
(±0.70)
4.36
(±0.53)
3.72
(±0.46)
3.66
(±0.64)
11
Suplementary material
29
Table 3. Direct comparison of IASI+GOME-2 ozone retrievals against raw ozonesonde 1
measurements. The same as Table 2 but without smoothing ozonesonde profiles with 2
IASI+GOME-2 averaging kernels. We also provide estimations of the a posteriori total and 3
smoothing errors, totcols and scol
smooth , for the IASI+GOME-2 retrievals used in the comparison. 4
5
Direct comparison of IASI+GOME-2 ozone retrieval vs. Raw ozonesondes
Atmospheric
column
Bias RMS R totcols scol
smooth
LMT -0.1
(-0.7 %)
2.4
(24.2 %)
0.75 2.0
(20.2 %)
1.8
(18.0 %)
LT -0.7
(-3.2 %)
4.4
(20.4 %)
0.58 2.6
(12.2 %)
2.2
(10.3 %)
TROPO -2.0
(-4.5 %)
6.9
(15.6 %)
0.86 3.8
(9.0 %)
3.3
(7.7 %)
UPTO30 -4.7
(-1.8 %)
15.4
(5.7 %)
0.92 7.4
(2.9 %)
6.2
(2.5 %)
6
Suplementary material
30
Table 4. Evaluation of potential effects of spectroscopic incoherencies between the UV and 1
TIR datasets on the multispectral ozone retrieval. The same comparison (119 ozonesondes 2
smoothed by the retrieval AVKs) as in Table 2 is performed, but for 2 test cases: (left) either 3
using TIR spectroscopic parameters from HITRAN 2000 or (right) decreasing UV cross-4
sections by -5 % with respect to BRION1993. 5
6
7
IASI+GOME-2 with HITRAN 2000 / with -5% UV cross sections vs. Ozonesondes
Atmospheric column Bias RMS R
LMT -1.6 / -1.2
(-14.7 % / -12.0 %)
2.7 / 2.4
(24.9 % / 22.8 %)
0.69 / 0.78
LT -4.3 / -2.9
(-18.5 % / -13.0 %)
6.2 / 5.1
(26.8 % / 22.9 %)
0.40 / 0.54
TROPO -5.4 / -4.4
(-10.9 % / -9.6 %)
8.7 / 7.7
(17.4 % / 16.4 %)
0.85 / 0.88
UPTO30 -6.6 / 4.0
(-2.5 % / 1.5 %)
16.6 / 14.8
(6.2 % / 5.7 %)
0.91 / 0.93
Suplementary material
Top Related